Overview of AmberClassic

• This repository contains msander, a "modern" version of parts of the Amber molecular dynamics program sander. Also included are various NMR, X-ray and cryoEM-related code and utilities, as well as versions of a number of the “classic” (and commonly-used) parts of AmberTools: tleap, antechamber, sqm, NAB, nabc, metatwist, rism1d, saxs, gbnsr6, xtalutil and paramfit. All of the force field files from AmberTools are also included here. With these tools, many systems can be set up for simulation in msander.

• The documentation and authorship credits are in the doc/AmberClassic.pdf file.

• This code is probably most useful to those who are already familiar with AmberTools (https://ambermd.org). Many of the basic tutorials there will also work with AmberClassic -- see the doc/AmberClassic.pdf for details.

• This package may also be of interest to those who want just the subset included here of the far-more-complex AmberTools package. Some other popular parts of AmberTools are not included here, but are available separately: cpptraj and pytraj (both at https://github.com/Amber-MD), and parmed (at https://github.com/ParmEd).

Warning

• This is a work in progress, essentially creating a (modified) subset of AmberTools. Please create a github issue if you have comments or suggestions. (As an alternative, send email to [email protected].) Volunteers to help improve or extend the package are welcome.

Design goals

• This project began as a fork of the sander code in AmberTools. It tries to (greatly) simplify the code base, choosing the best and most useful parts of the code, and to serve as a test bed for how modern Fortran coding techniques can be used. Key application areas are expected to be in structure refinements using NMR, cryoEM or Xray diffraction information. This version has a fair amount of OpenMP support, especially for Xray and 3D-RISM calculations. Parts of the Xray code use GPU acceleration.

• One additional goal of this collection is to make compiling and installation as simple as possible. There is a pretty simple configure script, and minimal dependencies on external packages. I am (slowly) cleaning up and adding other parts of AmberTools, and a conda package is available (see below.)

• This project incorporates and supercedes two previous packages (msander and nabc) that were available at github.com/dacase.

Key differences in functionality versus sander

• Some pieces are missing from the sander program in AmberTools:

  • Things that should be easy to re-introduce later: emil, sebomd, pbsa, APBS

  • Things are are problably gone for good, but which don't represent the best current practice: Path-integral methods, thermostats that don't follow the "middle" scheme, Berendsen barostat

  • Things that might be useful, but really complicate the code: evb potentials, QM/MM, nudged elastic band, constant pH and constant redox potential simulations. The API interface has also been removed.

  • Non-periodic 3D-RISM has been removed for now, in an attempt to get the simplest possible RISM code, perhaps as a basis for future GPU work.

(If you need some of these deleted pieces, use sander from AmberTools instead.)

• Key pieces of code that are still there, and being emphasized:

  • Periodic and non-periodic simulations, with all of Amber's GB models

  • 3D-RISM in periodic boundary conditions

  • NMR, cryoEM and Xray restraints (including quite a bit of new code; Xray restraints include NVIDIA GPU-enabled capabilities)

  • Thermodynamic integration and non-equilibrium sampling methods, including adaptively biassed sampling and self-guided Langevin dynamics

  • Sampling and minimization using the lmod and xmin approaches; these can now be used in conjunction with SHAKE and SETTLE.

  • Replica exchange capabilities, except for constant pH and redox potential simulations

Execution speed

Force field evaluation is still slow compared to many other codes. This project thus focusses on systems where other parts of the simulation (such as RISM or Xray/Cryoem restraints) are the bottleneck, so that force field speed is not the limiting component.

Building the code

• Compiling, on MacOSX, Linux, probably WSL:

   ./configure --help   #  then re-run configure with the options (if any) you want
   make install
   source AmberClassic.sh
   make test

See the doc/AmberClassic.pdf file for more details.

• Installing pre-built executables via conda

   conda create --name AmberClassic   # you can choose the env name
   conda activate AmberClassic
   conda install dacase::amberclassic  # only linux-64 and osx-64 for now
   cd $CONDA_PREFIX
   conda install -c conda-forge parmed # optional: brings in ParmEd
   source AmberClassic.sh
   .... # go to some work directory and have fun!
   conda deactivate  # to exit the conda environment when done

Again, see the doc/AmberClassic.pdf file for more details.

License

This project is licensed under the GNU General Public License, version 2, or (at your option) any later version. Some components use different, but compatible, open source licenses. See the LICENSE file for more information.